Comparing undergraduate learning outcomes and experiences across code-based and non-code-based statistical software...

This study examines whether undergraduate students in an introductory statistics course report different learning outcomes and experiences based on whether they use code-based or non-code-based statistical software. The sample included 2,241 students enrolled in courses using the Passion-Driven Statistics curriculum across 61 post-secondary institutions. Seventy-two percent of participants learned a code-based platform (R, SAS, Stata, Python), while 27.6% learned a non-code-based platform (SPSS, Excel, JMP, StatCrunch). Mixed-effects cumulative logit and logistic regression models were used to compare outcomes between groups while accounting for clustering of students within courses and adjusting for student demographics and academic background. Students in code-based courses had higher odds of reporting that they worked harder and found the course more challenging than those in non-code-based courses. At the same time, learning a code-based platform was positively associated with perceived gains in analyzing data for patterns, greater excitement about learning new concepts, and increased interest in conducting research. However, students learning code-based software reported feeling less prepared for advanced disciplinary coursework or thesis work. Overall, the results suggest that learning to work with code is associated with greater engagement and interest in data-driven work, even as it introduces greater challenges. These findings highlight the potential value of incorporating code-based statistical tools into undergraduate curricula while also underscoring the importance of supporting students through the initial learning curve.